Conclusions

In this Chapter, it has been described how NN can be used to create a nonlinear mapping between non-subjective video signals measures (i.e., packet loss rate, loss distribution, bit rate, frame rate, encoded frame type, etc.), and a subjective (i.e., MOS) measure of video quality. This mapping mimics the way in which human subjects perceive video quality at a destination point in a communication network. We have validated our approach by building a NN to assess in real time the video quality transmitted over the Internet, taking into account the previously mentioned parameters. We have shown that the NN performs quite well in measuring video quality in real time. Using the results of this Chapter, we present in the next Chapter a study of the impact of the mentioned parameters on video quality. We have implemented an application to measure in real time video quality in packet networks. This application incorporate an H263 encoder, a decoder and a network transmission simulator. With the advances in current processors and network bandwidths, it is possible to use software video codecs in real time (traditionally, the encoder required dedicated hardware to run in real time). Our application can be modified to implement a software application supporting real-time encoding, transmission, decoding, and measurement of video quality. One attractive use is video teleconferencing over packet networks. ChapterChapter